Hi! I'm a post-doc at Stanford, advised by Shreyas Vasanawala and John Pauly. My research interests are in signal processing, medical imaging, data science and inverse problems.

I did my undergrad and PhD at UC Berkeley, where I had the pleasure to work with Miki Lustig for many years. I have worked on sparse representations, fast MRI reconstruction algorithms, and open-source software packages.

In my Berkeley days, I have also collaborated with Kannan Ramchandran on sparse FFT algorithms. And I did a wonderful summer internship with Peyman Milanfar on kernel methods for image processing.

Large-scale 3D Dynamic MRI from Continuous Acquisitions

with Xucheng Zhu, Joseph Cheng, Kevin Johnson, Peder Larson, Shreyas Vasanawala, and Miki Lustig

3D dynamic MRI on the order of 100GBs can be reconstructed from continuous acquisitions using a compressed low rank representation and a stochastic algorithm.

Accelerating Non-Cartesian MRI Reconstruction with k-space Preconditioning

with Martin Uecker, and Miki Lustig

Non-Cartesian MRI reconstructions can be efficiently accelerated by preconditioning the convex dual problem.

Approximating Proximal Operators with Kernel Methods

with Peyman Milanfar, and Pascal Getreuer

Proximal operators can be approximated with kernel methods with appropriate regularization functions.

General Phase Regularized MRI Reconstruction with Phase Cycling

with Joseph Cheng, and Miki Lustig

Robustness against phase wraps in iterative reconstruction can be achieved by spatially shifting the wraps over iterations.

Multiscale Low Rank Matrix Decomposition

with Miki Lustig

Decomposing a matrix into block-wise low rank matrices of different scales can be done with a convex program and works well for many imaging tasks.

2D Sparse Fast Fourier Transform with Sparse Graph Codes

with Samir Pawar, and Kannan Ramchandran

2D DFTs of k-sparse spectrums can be computed in O(k log k) time by subsampling the signal with co-prime factors, e.g. 5, 7, and 8.

Robust 4D Flow Denoising with Divergence-Free Wavelet Transform

with Martin Uecker, Umar Tariq, Albert Hsiao, Marc Alley, Shreyas Vasanawala, and Miki Lustig

4D flow data can be robustly denoised by thresholding divergence-free wavelet coefficients.